MIT and Recursion Release Boltz-2: Next Generation AI Model to Predict Binding Affinity at Unprecedented Speed, Scale, and Accuracy
- Achieves near-FEP accuracy while being 1000x faster and less computationally expensive
- Outperforms all CASP16 affinity challenge participants in predictive power
- Open-source availability under MIT license for both academic and commercial use
- Trained on extensive dataset including 5 million binding affinity assay measurements
- Unique capability to model 3D complex structures while predicting binding affinity and protein dynamics
- None.
Insights
Recursion's Boltz-2 model dramatically accelerates drug candidate screening, potentially transforming early-stage discovery economics while maintaining accuracy.
Recursion and MIT's new Boltz-2 model represents a significant technological breakthrough in the computational drug discovery space. The model simultaneously predicts both molecular structure and binding affinity—approaching the gold-standard accuracy of free energy perturbation (FEP) calculations but at 1000x faster speeds. This acceleration directly addresses one of the most persistent bottlenecks in early drug discovery: efficiently triaging vast compound libraries to identify promising candidates.
The technical advantages are substantial. Unlike previous models that focused primarily on structure prediction (like AlphaFold), Boltz-2 performs co-folding that simultaneously assesses binding strength. This dual capability transforms the economics of virtual screening campaigns by maintaining FEP-like accuracy while dramatically reducing computational costs. The model has demonstrated superior performance in the CASP16 affinity challenge, validating its predictive capabilities against established benchmarks.
What makes this particularly valuable is the open-source nature of the release—the model, weights, and training pipeline will be available under an MIT license for both academic and commercial applications. This could accelerate adoption across the industry and potentially democratize access to sophisticated computational screening that was previously cost-prohibitive for smaller organizations.
The training data scope is also noteworthy, including approximately 5 million binding affinity assay measurements, molecular dynamics simulations, and expanded distillation data. This robust training foundation likely contributes to the model's performance advantages over existing approaches.
Recursion's Boltz-2 platform significantly enhances its competitive position in AI-driven drug discovery while strategically expanding its technological moat.
This MIT collaboration substantially strengthens Recursion's position in the competitive AI-powered drug discovery landscape. By developing and open-sourcing a model that dramatically outperforms existing methods in both speed and accuracy, Recursion establishes itself as a technological leader while simultaneously building goodwill within the scientific community.
The strategic decision to open-source this breakthrough technology might seem counterintuitive, but it represents sophisticated platform thinking. By making the base model widely available, Recursion accelerates the field's advancement while retaining proprietary advantages through its integration with their comprehensive BioHive-2 infrastructure and likely proprietary training datasets. This approach creates network effects that benefit Recursion disproportionately compared to competitors.
From a commercial perspective, Boltz-2's capabilities address a critical inefficiency in drug discovery economics. The model's ability to screen compounds at near-FEP accuracy but 1000x faster potentially transforms the economics of early discovery by reducing false positives that waste downstream resources. For context, traditional high-throughput screening can cost
The NVIDIA-powered computing infrastructure mentioned (BioHive-2) represents another competitive advantage, as computing capacity remains a significant barrier to entry in this field. This release demonstrates Recursion's ability to leverage this infrastructure investment to produce tangible technological breakthroughs, validating their capital allocation strategy.
- Boltz-2 is the first biomolecular co-folding model to combine structure and binding affinity prediction, approaching the accuracy of physics-based free energy perturbation (FEP) calculations but at speeds up to 1000x faster in standard benchmarks
- The development of this open source model for academic and commercial use was a collaborative effort, combining MIT’s deep academic expertise with Recursion's AI research and NVIDIA-accelerated supercomputer, BioHive-2
Salt Lake City, UT, June 06, 2025 (GLOBE NEWSWIRE) -- Researchers at the Massachusetts Institute of Technology (MIT) Computer Science and Artificial Intelligence Lab (CSAIL) and Jameel Clinic, alongside TechBio company Recursion (NASDAQ: RXRX), today announced the open-source release of Boltz-2, a first of its kind biomolecular foundation model. Powered by Recursion's NVIDIA supercomputer for its training and validation, this next-generation AI model achieves best-in-class accuracy in jointly modeling complex structures and binding affinities. Boltz-2 represents the next step beyond existing biomolecular structure prediction models like AlphaFold3 and its predecessor, Boltz-1.
“Accurately predicting how strongly molecules bind has been a long-standing challenge in drug discovery—one that required novel machine learning and computer science techniques to address,” said Regina Barzilay, MIT School of Engineering Distinguished Professor for AI and Health, AI faculty lead at Jameel Clinic and CSAIL principal investigator. “Boltz-2 not only addresses this crucial problem but also helps scientists uncover new biological insights and ask questions they couldn't before with standard approaches that are more computationally intensive. Because Boltz-2 is open-source, including its training code, scientists can easily adapt it for specific types of molecules, making it even more powerful as a tool to accelerate discovery."
Specifically, Boltz-2 marks a new era for in silico screening, in standard benchmarks approaching the accuracy of physics-based free energy perturbation (FEP), an industry-standard computational method used to predict the binding affinity of molecules, at speeds up to 1000x faster. The decrease in cost and increase in speed and scale makes large-scale and accurate virtual screening more practical than previously possible, directly addressing a critical bottleneck in small molecule discovery.
"Selecting the right molecules early is one of the most fundamental challenges in drug discovery, with implications for whether R&D programs succeed or fail," said Najat Khan, Chief R&D Officer and Chief Commercial Officer at Recursion. “By predicting both molecular structure and binding affinity simultaneously with unprecedented speed and scale, Boltz-2 gives R&D teams a powerful tool to triage more effectively and focus resources on the most promising compounds. Collaborations like this, bridging academic innovation and industry application, play an important role in advancing the field and, ultimately, improving how we develop and deliver medicines for patients."
Below are key components and differentiators of Boltz-2 vs other methods of predicting biomolecular structures and affinities:
- Improved Affinity Prediction: Near-FEP accuracy on the widely adopted FEP+ benchmark while being over 1,000 times faster and less computationally expensive
- Leading Benchmark Performance: Superior predictive power, demonstrating outperformance over all CASP16 affinity challenge participants
- Advanced Joint Modeling: Uniquely models 3D complex structures while jointly predicting binding affinity and protein dynamics (e.g., B-factors)
- Controllable & Physically Realistic: Achieves significantly improved physical plausibility using Boltz-steering and offers enhanced user control via template, method, and contact conditioning
- Novel & Expanded Training Data: Trained on molecular dynamics simulations, expanded distillation data, and approximately 5 million binding affinity assay measurements
In line with MIT and Recursion’s commitment to making AI tools accessible for drug developers, Boltz-2 will be open-sourced under an MIT license, making the model, weights, and training pipeline available for both academic and commercial use.
Boltz-2’s development was led by the Boltz team at MIT under the supervision of Professors Regina Barzilay and Tommi Jaakkola alongside a team of researchers from MIT and Recursion. For more information, visit: https://boltz.bio/boltz2.
About Recursion
Recursion (NASDAQ: RXRX) is a clinical stage TechBio company leading the space by decoding biology to radically improve lives. Enabling its mission is the Recursion OS, a platform built across diverse technologies that continuously generate one of the world’s largest proprietary biological and chemical datasets. Recursion leverages sophisticated machine-learning algorithms to distill from its dataset a collection of trillions of searchable relationships across biology and chemistry unconstrained by human bias. By commanding massive experimental scale — up to millions of wet lab experiments weekly — and massive computational scale — owning and operating one of the most powerful supercomputers in the world, Recursion is uniting technology, biology and chemistry to advance the future of medicine.
Recursion is headquartered in Salt Lake City, where it is a founding member of BioHive, the Utah life sciences industry collective. Recursion also has offices in Toronto, Montréal, New York, London, Oxford area, and the San Francisco Bay area. Learn more at www.Recursion.com, or connect on X (formerly Twitter) and LinkedIn.
Investor Contact Investor@Recursion.com Media Contact Media@Recursion.com
FAQ
What is Boltz-2 and how does it improve drug discovery?
How does Boltz-2 compare to existing methods like FEP?
What are the key features of RXRX's Boltz-2 model?
Who can use Boltz-2 and how is it licensed?
Who developed Boltz-2 and what technology was used?
